Security audit

宏观脉搏

Security checks for vulnerabilities and agentic risk

Overview

This macro-reporting skill is mostly purpose-aligned, but it needs review because it combines scheduled runs, local writes, outbound report delivery, and broad modification/reset wording without tight scoping.

Install only if you want an automated macroeconomic monitoring agent. Before enabling the cron schedule or webhook/email delivery, confirm where reports are sent, what local files it may update, and ignore or constrain the generic delete/reset/import capability language unless the publisher documents exact safe commands.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (6)

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The troubleshooting section instructs the agent to run network diagnostics such as ping and to inspect firewall/proxy settings, which exceeds the stated purpose of macroeconomic data monitoring and reporting. In an agent context, unrelated diagnostic actions can expand operational scope, trigger unintended network activity, and normalize command execution outside user intent.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The skill introduces broad claims for create/query/modify/delete/export/reset/import operations that are not supported by the documented macro-monitoring workflow. Overbroad capability language can cause an agent to interpret routine requests as authorization for state-changing actions, increasing the risk of unintended file modification, data export, or configuration resets.

Intent-Code Divergence

Medium

Confidence: 91% confidence
Finding: The same section claims support for create/modify/delete-style modes even though the rest of the file describes a read-heavy monitoring and reporting workflow. This contradiction creates unsafe ambiguity about permitted actions and can lead an agent to perform destructive or persistent operations not reasonably expected for the skill's purpose.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The skill metadata uses very broad 'Use when' language covering generic data analysis, report generation, statistical insight, and visualization. This can cause over-triggering in unrelated user requests, making the agent invoke file writes, web access, or outbound messaging when the user only wanted general analysis.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The manual trigger is expressed as ambiguous natural language rather than a clearly delimited command or schema. In agent systems, vague triggers increase the chance of accidental activation and downstream side effects such as browsing sources, writing files, and sending outbound messages without sufficiently specific user intent.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill instructs writing to local files and sending reports through IM, email, or webhook channels but does not prominently warn users about persistence and external data transmission. Without explicit disclosure and consent, the agent may store content or exfiltrate summaries to third-party endpoints in ways the user did not anticipate.

Static analysis

No suspicious patterns detected.